Development and validation of a model incorporating choroid plexus morphology from brain MRI to predict anti-TNF-α therapy outcomes in Crohn’s disease

Central nervous system involvement has been implicated in the pathophysiology of Crohn’s disease (CD). This study aimed to develop a predictive model integrating choroid plexus (CP) morphology derived from brain MRI with clinical and psychological factors to forecast treatment outcomes following anti-tumor necrosis factor-alpha (anti-TNF-ɑ) therapy in CD. This prospective study enrolled adult patients with CD who underwent brain MRI prior to initiating anti-TNF-ɑ therapy between 2019 and 2022. All participants received brain MRI scans within two weeks of study inclusion and before treatment commencement. The CP was manually segmented from baseline MRI scans, and mesh volume (MV) and surface area (SA) were calculated. Patients were evaluated for short-term (12week) and long-term (≥ 48 week) clinical remission (STCR and LTCR, respectively). A total of 130 patients were included, of whom 73% received infliximab and 27% received adalimumab. Both MV and SA of the CP correlated with systemic inflammatory markers and psychological symptom scores. Multivariate logistic regression identified hemoglobin level, emotional function score, and SA of CP as independent predictors of STCR, while MV of CP emerged as the sole independent predictor of LTCR. Combined models incorporating these variables demonstrated excellent predictive performance, with areas under the curve of 0.895 for STCR and 0.939 for LTCR. This study identifies CP morphology as a novel, non-invasive neuroimaging biomarker. A model integrating CP characteristics with clinical parameters accurately predicts response to anti-TNF-ɑ therapy in patients with CD. We developed an integrated predictive model combining choroid plexus morphology with clinical parameters. Although individual choroid plexus features showed only modest correlations with clinical and psychological factors, the combined model exhibited robust predictive accuracy for both short-term and long-term clinical response to anti-TNF-ɑ therapy in patients with Crohn’s disease.